Asiya is the machine translation evaluation toolkit to score machine translation outputs (http://asiya.lsi.upc.edu/). It is largely written in Perl.
How do I use Asiya to perform BLEU metrics?
I have followed the youtube introduction video: https://www.youtube.com/watch?v=rA5De9Z4uWI
And created a config file (Asiya.config):
input=raw
srclang=en
srccase=en
trglang=ja
trgcase=ja
src=corpus.tok/test.tok.en
ref=corpus.tok/hyp.tok.ja
sys=corpus.tok/test.tok.ja
some_metrics= BLEU NIST METEOR-ex Ol
My machine translation output file is in corpus.tok/hyp.tok.ja
, the source file is in corpus.tok/test.tok.en
and the reference file (correct translation) is at corpus.tok/test.tok.ja
. They are tokenized plain text files, each line is a sentence.
And when I ran:
/home/expert/asiya/bin/Asiya.pl -eval single -g all -metric_set some_metrics Asiya.config
I got this error:
Smartmatch is experimental at /home/expert/asiya/bin/../lib/IQ/Common.pm line 784.
Smartmatch is experimental at /home/expert/asiya/bin/../lib/IQ/Common.pm line 791.
Smartmatch is experimental at /home/expert/asiya/bin/../lib/IQ/Scoring/ESA.pm line 339.
Use of uninitialized value in concatenation (.) or string at /home/expert/asiya/bin/../lib/IQ/Config.pm line 251.
[ASIYA] directory </tools> does not exist!
The tools directory does exists as a subdirectory in the current directory that I ran the command. What went wrong? Is there a parameter that I could add for the Asiya tools directory?
How do I use Asiya to perform BLEU evaluation?
If I don't use Asiya, how else can I get the BLEU score per sentence and system BLEU scores for my machine translation output?
(more details on http://nlp.lsi.upc.edu/redmine/boards/11/topics/138)